Intelligent Logistics Path Optimization Algorithm Based on Internet of Things Sensing Technology

Abstract

This paper studies the logistics path optimization problem based on the Internet of Things (IoT) and deep learning, and proposes a hybrid algorithm (DRL-GA) that integrates deep reinforcement learning (DRL) and genetic algorithm (GA). Through sensors, RFID tags and other devices installed in vehicles, goods and warehouses, logistics data is collected in real time and transmitted to the cloud through wireless communication technology for big data analysis. The DRL model dynamically adjusts the path selection using real-time data, while the GA optimization module performs a global search on the paths generated by DRL to ensure the optimality of the path. Experimental results show that the DRL-GA hybrid algorithm significantly outperforms other baseline methods in key indicators such as total path cost, computation time, convergence speed and solution quality, especially when processing large-scale data sets. In addition, the algorithm also shows good adaptability and stability in robustness tests under different environments. Experimental results show that the DRL-GA hybrid algorithm significantly outperforms other benchmark methods in key indicators such as total path cost, computation time, convergence speed and solution quality." It was then added that "On the small-scale dataset Eil51, compared with the genetic algorithm, the DRL-GA hybrid algorithm reduced the computation time by 0.01 seconds, increased the convergence speed by 9 iterations, and narrowed the gap between the solution quality and the optimal solution by 0.008%. On the medium-scale dataset Ch150, the total path cost was reduced by 56.7 and the computation time was shortened by 0.03 seconds. These quantitative results fully demonstrate the superiority of the hybrid algorithm.